How AI Diagnostic Tools Can Lead to Financial Burdens for Patients

How AI Diagnostic Tools Can Lead to Financial Burdens for Patients

According to the co-authors of a perspective published in the New England Journal of Medicine, the use of AI diagnostic tools by medical practitioners can lead to unexpected financial burdens for their patients.
AI Diagnostic Tool-Illustration Getty Images

This brief was initially published by Stanford Law School.

In this New England Journal of Medicine perspective, Stanford Health Policy's Michelle Mello, JD, PhD, professor of health policy and of law, looks at what happens when AI tools recommend confirmatory diagnostic testing, but insurance companies won’t cover the tests. According to the perspective by Mello and her Stanford Medicine co-authors Nigam H. Shah and Sneha S. Jain:

  • Some AI-based health care tools are used to identify patients who are at risk for disease or other adverse health outcomes and may benefit from diagnostic testing, monitoring, or treatment that they would not otherwise have received.
  • Insurance coverage of confirmatory diagnostic testing (to ensure the patient truly has a medical condition) is usually available when a patient identified in this way has a medical history that justifies receipt of confirmatory diagnostic testing under the current standard of care.
  • Challenges arise when identified patients lack conventional risk factors or symptoms, and patients may not receive insurance coverage for these confirmatory tests.
  • Although identification of patients who are at risk but are missed by traditional risk assessments is an important part of the value proposition for using AI tools, insurance coverage may not be readily granted for such patients even if AI tools are appropriately validated.
  • Only 37% of adults in the United States can pay an unexpected expense of $400 without borrowing money or selling an asset, and so if patients are denied coverage of confirmatory diagnostics, this may exacerbate health care disparities between high-income and low-income patients (those who can pay for a confirmatory test and those who cannot).
  • The potential benefits of AI use in health care are thus inequitably distributed.
  • There are several steps healthcare systems deploying AI should take to protect their patients from potential financial toxicity of AI use, and there is an urgent need for insurers to update their processes around coverage determination.

     

The full paper can be read here. (NEJM account required)